Merge commit 'refs/top-bases/linux-have-uio-rw' into linux-have-uio-rw

This commit is contained in:
Brian Behlendorf 2009-02-04 16:31:06 -08:00
commit babb8a5b66
1 changed files with 26 additions and 11 deletions

37
TODO
View File

@ -1,16 +1,5 @@
SUMMARY OF MAJOR KNOWN PROBLEMS IN v0.4.1 (Development Release)
- SPLError: 10167:1968:(spl-kmem.c:1286:spl_kmem_cache_reap_now())
ASSERTION(skc->skc_magic == SKC_MAGIC) failed
The above assertion is overserved when perform more IO than can be fully
cached by the ARC. The Linux VM applies back pressure to the slab which
in turn detect what appears to be memory corruption. I've seen a few
flavors of this so far, so I'm not yet convinced this is actually an
issue with the SPL slab. It may just be the most common victim, more
investigation is needed. It is also possible the new untested vdev_disk.c
is to blame. A lot of work is needed here.
- 2009-01-26 16:23:00 SPLError: 7884:1360:(dnode.c:146:dnode_verify())
VERIFY3(dn->dn_nblkptr >= 1) failed (0 >= 1)
@ -20,3 +9,29 @@ SUMMARY OF MAJOR KNOWN PROBLEMS IN v0.4.1 (Development Release)
disabled until it could be explained. Well it has not just gone
always with the update to b105 so we need to run it to ground and
explain what is going on.
- Implement FLUSH support probably in terms of a barrier as part of
the vdev_disk.c interface. Required for correctness.
- Implement something similar to the Solaris devid support to ensure
ZFS properly locates your disks even when moved. The current port
is dependant on using something like udev to ensure this can never
happen but this is not a viable long term solution.
- Implement something like Solaris's sysevent support to detect when
drives appear or are removed. There is no easy analog to my knowledge
for linux and we will need to come up with something.
- Integrate Ricardo XDR support for on disk encoding. This mainly
just needs a good review and to be updated against this code tree.
- Ztest failures have been observed after 8+ hours of runtime. This
needs to be investigated and resolved.
- Get the ZVOL working as a vehicle for further stress testing under
Linux, and to provide one useful user space interface for access to
the DMU.
- Get the ZPL working minimal support will be needed for lustre.
- If Richardo is willing integrate his FUSE port in to this code base.